Objectionable taste and odour in water supplies in North-East ...

DRINKING WATER INSPECTORATEArea 4AErgon House,Horseferry RoadLondon SW1P 2ALEnquiries: 0300 068 6400Facsimile: 0300 068 6401E-mail: marcus.rink@defra.gsi.gov.ukDWI Website: http://www.dwi.gov.ukRef.2010-2661/266325 October 2011Objectionable taste and odour in watersupplies in North-East London betweenJanuary and March 2010Marcus RinkDeputy Chief Inspector (Operations)1

CONTENTS1. Executive Summary 32. Recommendations 73. Introduction 94. Taste and Odour 115. Identification of the Contaminants 156. Source of the Contaminants 177. Raw Water Management 218. Final Water Management 259. Health Authority Liaison 2710. Consumer Perceptions 2911. Media 3312. Liaison with External Stakeholders 3413. Lessons Learnt 3514. Contraventions of the Water Supply (Water Quality) Regulations 3615. Notification 3716. Water unfit for human consumption 3817. References 39Appendix I: Raw Water Supply Schematic 40Appendix II: Water Treatment and Supply Schematics 41Appendix III: Abstraction quantities and supply percentages of raw water 43Appendix IV: Company contact graphs 45Appendix V: Consumer contact distribution maps 47Appendix VI Method synopsis of 2-EMD/2-EDD 49Appendix VII: Technical details of 2-EDD & 2-EMD 52Appendix VIII: Customer Questionnaire Responses to DWI 57Appendix IX: BBC Website reports of event 592

1. Executive Summary1.1. This reports sets out the conclusions and recommendations arising fromthe Inspectorate‟s assessment of an event where there was anobjectionable taste and odour in water supplied from three treatmentworks (Coppermills, Chingford South and Chigwell) serving the NorthEast London area and occurring in January through to March 2010. Intotal there were 1,114 recorded consumer contacts about the event toThames Water and a further 97 recorded consumer contacts to Essexand Suffolk Water.1.2. When notified of an event, the Inspectorate gathers informationconsidered to be relevant and assesses this in conjunction withinformation provided by the company about the circumstances of theevent and any actions taken. The Inspectorate then considers the wayin which the event was handled and whether any breaches of regulatoryduty or offences occurred. Thames Water notified the Inspectorate ofthis event on 09 February 2010 and notification by Essex and SuffolkWater was on 11 February 2010.1.3. This event affected supplies managed by two water companies; onecompany because it relies on a source of raw water provided as acommercial bulk supply by the other and the other because it both usesand exports the same raw water source. The common causal factor inthe event was the raw water abstracted from the River Lee into acomplex series of surface water storage reservoirs.1.4. About 10 miles upstream of the intake to the reservoirs is an outfallwhere final effluent from Rye Meads Sewage Treatment Works, (STW)is discharged. The STW has three treatment lines, one dedicated for useby a commercial licensed waste company. It was this commercial wasteline which was found, on investigation by Thames Water, to containwaste from a resin manufacturer that contained the two organicsubstances which caused the objectionable taste and odour detected byconsumers in North East London during January and February 2010.The same chemicals have been implicated in similar previous eventsworldwide (Pennsylvania, Barcelona and Worcestershire) during the1990‟s. This report has been prepared to add to the body of publishedknowledge about the risks to drinking water posed by disposal of thistype of waste material to the aquatic environment.1.5. Two surface water storage reservoirs (King George V and WilliamGirling) serve the Chingford South Water Treatment Works (WTW),owned by Thames Water Utilities Limited. The William Girling suppliesChigwell WTW owned by Essex and Suffolk Water a subsidiary ofNorthumbrian Water Limited. Together with the Lee Valley Complexreservoirs these reservoirs also serve Coppermills WTW owned byThames Water.1.6. An important contributory factor in this event was the particularconditions of low dilution of the sewage effluent in the small riverreceiving the discharge. A further confounding factor was essentialmaintenance work taking place on the embankment of the King GeorgeV. This reservoir receives 70% of its water from a point located upstream3

of the STW. This water is derived from the River Thames and transferredto the Lee Valley by the Thames Lee raw water tunnel. Under normalcircumstances and prior to January 2010 the reservoir contributed onaverage 8% of the water treated at Coppermills WTW. The embankmentworks commenced in the winter of 2009 and required Thames Water toreduce the volume of stored water in the reservoir by minimising draw offor closing the intake during December 2009 through to January 2010.On the week beginning 8 January the usage of King George V went upby around 10% to 17%, (maximum increase in usage during event wasby 13% from 7% to 20%), at a time when the intake was closed and thedemand on the Thames Lee Tunnel fell by 6%, (maximum reductionduring event of 21% from 32% to 11% contribution). This change inoperating regime to facilitate embankment works is consideredretrospectively to mark the onset of the event with the first peak ofcustomer complaints being received by Thames Water in a hydraulicallyconsistent time period (around a week later). The operating regimeinadvertently had the effect of increasing the concentration ofcontaminants in water entering the treatment works but it also identifiesthe point in time before which the contaminated waste entered RyeMeads STW, (before January 2010), as the contamination must havealready been in the reservoir.1.7. Looking at the Essex and Suffolk Water case, the single supply fromWilliam Girling reservoir of about 95 ML/day would have received thecontaminants direct from the River Lee, the estimated time to fullyturnover the William Girling reservoir is about forty days and recordedconsumer complaints began to be received from 3 February 2010.1.8. The population potentially affected was about 2.25 million Thames Watercustomers in North-East London and 0.5 million Essex and SuffolkWater customers in the North-East and East Greater London areaincluding Ilford, Barking, Dagenham and Romford1.9. Critical to the identification of the cause of the event was a telephone callmade by Essex and Suffolk to Thames Water on the morning of 11February 2010. This call was prompted by the company detecting astrong PVA glue and citrus smell at a consumer’s tap following a routineprogrammed sample and subsequent investigation of calls fromcustomers reporting similar unusual taste and odours in tap water. Thiscommunication connected the consumer contacts being received byboth companies and raised awareness of the two events having acommon cause which in turn pointed to this being the raw water quality.1.10. Thames Water acted to cease abstraction from the River Lee and beginusing the Lee Valley Complex reservoirs increasing the contribution fromthe Thames Lee Tunnel to 37% introducing uncontaminated RiverThames water into the supply system from 11 February onwards.1.11. Thames Water identified the chemicals 2-ethyl-5,5-dimethyl-1,3-dioxane(2-EDD) and 2-ethyl-4-methyl-1,3-dioxolane (2-EMD) in both the finaland raw water on 16 February 2010 at levels considered by the HealthProtection Agency not to be a risk to health. The identification was madeusing expertise available through the industry‟s mutual aid laboratorynetwork. The request to this network was key to identification of thecausative chemicals and was a commendable decision. In previous4

historical events, the identification of these chemicals has proved to beproblematic because they are small, polar and highly soluble whichmeans they are not identified through routine analysis and requirespecial methodology to be deployed. Delays in identification ofcontaminants were a significant criticism of the report into the RiverSevern Pollution (Wem) incident of 1994 3 and the establishment and useof a mutual aid network was an important recommendation to theindustry at that time.1.12. On 18 February sampling and analysis had linked the contamination inwater to a single commercial waste stream at Rye Meads (STW).1.13. The start of the event is defined by the change in consumer taste andodour reporting rather than by the contamination. Consumer contacts toThames Water increased through January. Towards the end of Januaryand into the first week of February the numbers rose significantly,coincident with the first complaint being received by Essex and SuffolkWater on 3 February which marked the beginning of the event for theirconsumers. Diagnosis of the significance of consumer contact data isgenerally an imprecise process because of the influence of other factors;for example, consumers have a known tendency to report qualityconcerns in response to receiving their water bill and in this event therewas an unrelated outbreak of viral diarrhoeal illness focusedgeographically on the catchment of Whipps Cross Hospital which iswithin the water supply zone of this event. Notwithstanding thesedifficulties, both companies were unable to detect the objectionable tasteand odour, despite it being discernible to consumers. This observationwas common to the investigation of the River Trent (Wem) Pollutionincident in 1994. The sensitivity of individuals to the contaminants, thetemperature of the water when analysed, the time the water remainsstanding and the place of the analysis, in particular the ventilation of thesurrounding area, all influence the ability to detect the odour. Neithercompany had optimised their taste and odour analysis or screened thepanellists for their ability to detect these contaminants.1.14. A further common finding with historical events was the lack ofawareness of these two chemicals by all relevant company staff. As aresult, Thames Water had failed to appropriately risk assess the hazardto enable mitigation and therefore did not have the information readilyavailable to understand what to look for and what actions to take.Consequently, the Thames Water investigation focussed on a number ofaspects whilst failing to consider the probable cause as part of the widerinvestigation. The absence of reliable information and the subsequentdecisions, to some degree, unavoidably extended the duration of thetaste and odour event.1.15. Essex and Suffolk Water relied on Thames Water for these assessmentsand had no water quality agreement in place to ensure that the rawwater provided was of suitable quality for treatment and secure promptcommunications and action about any potential raw water qualityproblems; consequently Essex and Suffolk Water were not aware theywere being supplied with contaminated raw water for a week.1.16. There are many similarities between this event and those which haveoccurred in the past but the learning from these events appears to have5

een lost largely because it appears not to have been embeddedeffectively in operational arrangements. This report thereforeemphasises these learning points and contains new considerations inthe context of today‟s water industry. Most important of these is the needto carry out a thorough risk assessment of waste streams including theidentification of every customer within that waste stream and whether ornot the waste originates from resin manufacturing. Equally watercompanies water safety plan approach should consider the specific riskspertaining to these chemicals and the wider effects that changes(temporary or permanent) in the configuration of raw water supplyarrangements may have upon that risk assessment.1.17. Wherever water is exported and imported, be it raw or treated water,there should be water quality agreements between importers andexporters and these should contain unambiguous accountabilities,contingencies, responsibilities, including the provision and sharing of riskassessments, and lines of communication in response should be set outin the event of any abnormal situations that may affect water quality.1.18. Critically, taste and odour methodology and personnel performing thistest on site lacked the capability to enable detection of the contaminantsand so both companies failed to detect the raw water contamination as itentered the treatment works. Without this, the companies did not havethe necessary information required to manage the circumstanceseffectively.1.19. This event was diagnosed by consumers contacting the companies andunfortunately for Thames Water, information from the water quality callhandling systems was not interpreted early enough to identify thebeginnings of a wide scale problem in a large water supply area; thishindered prompt identification and remediation of the cause and also ledto a loss in public confidence through untimely or inappropriateconsumer communications.6

2. Recommendations2.1. Risk assessments for waste streams must examine each and everyindividual contributor to that waste so that any potential hazard to watersupplies is identified. This should be the collective responsibility of thesewage treatment works operator and any water company drawing onthe raw water for water supply purposes. This must not be solely aresponsibility of the waste operator, commercial or otherwise, andinformation should not be withheld on the basis of commercialconfidentiality. In a situation where information is withheld, waste shouldnot be accepted. Where there are private water supply users relying onthe same raw water source water companies should share the riskinformation with local authorities responsible for private water supply riskassessments.2.2. All water and wastewater companies must ensure that any waste whichmaybe a source of 2-EDD and 2-EMD is closely monitored and any resinmanufacturer should be classified as high risk. Waste where the risk ishigh or unacceptable and cannot be mitigated should not be acceptedwhere it could subsequently enter a raw water source however remote.2.3. Any unusual odour discerned at a sewage treatment works should beacted upon and resolved, whether or not it is connected to an external orpublic complaint.2.4. Monitoring of sewage effluent, where a risk exists, should include 2-EDDand 2-EMD and this should be embedded in any contractual agreementin the acceptance of waste to a sewage treatment works.2.5. The effect (or not) of any dilution of effluent by the receiving river shouldbe risk assessed and understood in relation to its function in anydownstream water supply risk mitigation.2.6. Any planned or unplanned changes in raw water supply arrangements(abstraction, storage and blending) should be risk assessed in relation tothe assumptions inherent in the water supply risk assessments andmitigation measures being relied upon routinely by all abstractors (publicand private).2.7. Where a bulk supply exists, be this raw or final water, there should be: unambiguous water quality criteria of suitability and contingency arrangements when criteria are not met and sharing of all relevant information to enable effective riskassessment and mitigation by each party and clear lines of communication and accountability for timely actionin response to any change in the supply arrangements or anyunusual findings, bi-directionally.2.8. Companies should ensure that taste and odour methodology on site issufficiently robust to detect contaminants like 2-EDD and 2-EMD bypersonnel who are screened for their capacity to detect them. Thereshould be recognition of the variation amongst individuals in this regard7

and where detection on-site cannot be reliably achieved then there mustbe arrangements for routine testing at a controlled laboratory facility withpanellists who can meet these requirements.2.9. Companies should put in place operational odour testing for raw waterwhenever the risk assessment identifies an odour risk, observing theconditions of 2.8 and health and safety requirements.2.10. All water companies should support and maintain mutual aid analyticaland emergency networks to ensure knowledge of unusual contaminantsmaintained and kept up to date and regularly disseminated.2.11. Companies should invest in the development of methods of diagnosingand evaluating consumer contact data to improve upon the capacity forearly identification of an event in water supply areas which serve largepopulations or where water from a common source serves severaldifferent communities or more than one operational or administrativearea.2.12. Water companies should embed within their response plansconsideration of the early involvement of the Inspectorate in thedevelopment of communication strategies, particularly in relation topress releases and matters of wider public interest as well as anyinformation directed specifically to the needs of their customers.2.13. Water companies are reminded of their duty to provide all informationrelevant to the full assessment of an event by the Inspectorate.2.14. Water companies are reminded of their duties under the WaterCompanies Information Direction which requires timely notification of allevents potentially impacting on public confidence.8

3. Introduction3.1. Between 25 December 2009 and 31 January 2010, Thames Waterreceived 70 general water quality complaints from consumers within theNorth East London area. A further 71 complaints in the week followingand then 25 complaints in a single day on the 8 February 2010.Companies routinely receive consumer contacts and for Thames Waterthe typical number received for the area is about 10 per week. Whilst notexceptional during the first half of January, the number of contacts wasrising beyond this threshold level and in particular the composition ofcontacts being received were specifically about taste and odour, asopposed to the wider range of contact types normally reported in theWoodford and Finsbury Park area. The rising number of contacts abouttaste and odour should have been obvious to Thames Water in the firstweek of February. By contrast, Essex and Suffolk Water received theirfirst contact of an unusual taste and odour on 3 February 2010. A total of11 contacts were then received between 6 and 8 February 2010.Thames Water notified the Inspectorate on 9 February 2010 and alsoconsulted the Health Protection Agency (local unit). At this stage,Thames Water considered the taste and odour problem was probablyrelated to the chloramination process at Coppermills WTW and this wasthe basis of the initial conversation with the HPU. The Inspectoratenotes the omission in the company‟s report of the record of this first HPUconsultation and the company is reminded to ensure it keeps andprovides records of all communications with the HPA.3.2. A range of taste and odour contacts were received by Thames Waterconsumers through January 2010 with descriptions varying fromchlorinous, sewage, salty, onions and perfume. The company identified119 contacts received from 29 of the 42 Water Supply Zones (WSZ)downstream of Coppermills WTW between 03 January and 02 February2010 approximating to 3 or fewer contacts in any one zone in a singleday. None of these contacts were from consumers downstream ofChingford South WTW during January. However, overall the rise innumbers of contacts of a similar type was unusual and only an handfulwere described as a medicinal or TCP taste, most being classified as“other”. It is probable that the taste/odour descriptor type by thecustomer call centre masked the common factor for some time andmeant that the company did not identify the beginning of the taste andodour event expeditiously.3.3. Overall Thames Water collected more that 400 drinking water samplesand over a quarter were in response to consumer contacts. Amongconsumers interviewed by the Inspectorate, less than 10% from bothcompanies said they had been visited for sampling and none werecontacted before they had themselves contacted their water company.The Inspectorate notes the low response rate of the companies toconsumer contacts. This may have been a missed opportunity to identifythe cause at an early stage. Essex and Suffolk Water who, whilst unableto identify a problem at Chigwell WTW, was more responsive to9

consumer contacts quickly identified a strong unusual taste and odourdescribed as PVA glue and citrus about a week following receipt of thefirst known contact associated with the event.3.4. The first contact from a zone downstream of Chingford South WTW wason 08 February 2010. All Chingford South zones are also fed partially byCoppermills WTW.3.5. The first evidence of onsite investigation by Thames Water atCoppermills was 8 February 2010 when treatment works samplingincluded taste and odour. Thames Water was focusing on thechloramination process as the likely cause of the taste and odour and nounusual taste or odours were identified in on site tests at this time. Onthe 10 February 2010, Thames Water staff reported an unusual odour indrinking water at a North London workplace. This provided a clearopportunity for Thames Water to be aware of and investigate the eventfrom a different perspective.3.6. Essex and Suffolk Water took samples on 9 February 2010 from eachslow sand filter at Chigwell. Following overnight refrigeration andreanalysis of the taste and odour samples on 10 February 2010 anunusual odour was discerned. A routine consumer tap sample taken on10 February 2010 also failed the quantitative odour test with asignificantly high dilution number (DN) of 8 and a description of PVA glueand citrus.3.7. Essex and Suffolk Water on 11 February 2010 visited a number ofconsumers who had reported taste and odour. Three samples hadpositive quantitative dilution numbers with a PVA glue and citrusdescription.3.8. At 09:45 hrs on 11 February, Essex and Suffolk Water contactedThames Water as the bulk supplier of raw water. From this contact itbecame evident that Thames Water was receiving similar taste andodour contacts. This critical information led Thames Water to considerthat the taste and odour problem was associated with the raw watersupply.3.9. The Inspectorate noted how reliance on their own information andknowledge failed to lead Thames Water to determine a problem with theraw water supply up to this point.10

4. Taste and Odour4.1. Working on the basis that the raw water was the most probable cause,Thames Water stopped impounding water from the River Lee to theWilliam Girling, King George V and High Maynard reservoirs at 19:00 on11 February 2010 and contacted the Environment Agency (EA).However, under the raw water supply configuration, it was not possibleto change the bulk supply to Essex and Suffolk Water‟s Chigwell WTWfrom the William Girling. The bulk supply continued to be derived fromWilliam Girling storage reservoir from this point forward.4.2. Between 11 and 15 February Thames Water had detected nothinguntoward by analysis and was considering the probable cause asgeosmin or 2-methyl isoborneol, (MIB) as common natural taste causingcontaminants caused by Actinomycetes or Cyanobacteria, (blue-greenalgae). These organisms exhibit a very characteristic earthy or mustytaste and odour. Approximately 8% of the contacts to Thames Water butonly 2% to Essex and Suffolk described the taste and odour in this way.On review it may appear therefore that the company in taking the viewthat the cause was the chloramination process was reasonable. Thecompany had detected no odour or tastes within the upstream processlocations or in the raw water using on-site tests on 8 February. Howeverthis investigative strategy would prove to be a key failing in thediagnostic process as it diverted the company‟s attention away fromconsidering raw water as the root problem and led to the delay in theidentification of cause until contacted by Essex and Suffolk Water aboutsimilar contacts from consumers.4.3. Essex and Suffolk Water discerned a PVA glue and citrus taste andodour in seven consumer samples and one secondary filtrate samplefrom Chigwell WTW on 10, 11 & 12 February 2010. Taste and odourdescriptors used by consumers can be variable and should not be reliedupon wholly to direct investigations, however, the information providedby Essex and Suffolk Water‟s investigation was both specific and reliablehighlighting how Thames Water‟s failure to diagnose and analyse its ownconsumer contact information was a missed opportunity. In particular,the recording of the contacts as “other” rather than recording theconsumer description was unhelpful.4.4. Taste and odour assessment is considered a core operational test andskill. However, it is only semi quantitative and subjective by nature andits results have inherent uncertainty. This uncertainty is due todifferences of capacity of individuals to detect a taste and/or odourgenerally which can be exacerbated by other personal factors such asage, gender, health, smoking, preferred food type and recent foodconsumption. Individuals vary genetically in their balance of tastereceptors between the four types: sweet, sour, salty and bitter. Thiscoupled with the circumstances of when and where a sample is takenand the place of testing all contribute to uncertainty. Sensitivity to a giventaste or odour will differ according to the chemical, the temperature of11

the water and the environment such as tasting the water when outside,because of ventilation, and the vagaries of “grab” sampling which canmiss contamination that is either transient or varying in concentrationaround the threshold of detection.4.5. In the case of the particular contaminants in this event, they werepresent only in very low amounts (sub µg/l levels) in the drinking water,and they exhibit very low taste and odour thresholds so detection wasparticularly challenging. However in light of 166 consumer contacts byand on 8 February 2010, the action taken by Thames Water was to testtwo in process samples, one pre-disinfection and one post-disinfectionalongside three ex-works network taste and odour samples. There wereno further active investigational samples at the treatment works until the11 February 2010 when the company was promoted to consider a rawwater problem upon which investigational sampling for algal causesoccurred on four days between 11 and 15 February and on twooccasions included the Woodford Depot.4.6. The company at this point knew the problem was associated with theraw water from information provided by Essex and Suffolk Water and thedescriptors from customers were variably chlorinous, sewage, salty,onions and perfume, and this was added to by information from staff attheir own depot in Woodford. An algal problem would characteristicallyhave exhibited as an earthy, musty or woody odour. The interpretation ofthis data may have been a missed opportunity to focus effort. There area number of publications that discuss the characteristic taste and odourcauses and give a flavour profile analysis and two of these wereavailable to the water industry through the Inspectorate‟s web site andspecifically discuss taste and odours in water, consumer perceptions,methods and causes. The first is by the Standing Committee of Analysts:The determination of taste and odour in drinking waters 4 and the secondis entitled Factors causing off-taste in waters, and methods andpractices for the removal of off-taste and its causes 5 . Both describe arange of taste and odours, their origins and the chemical or biologicalcauses.4.7. The key diagnostic failing by Thames Water staff was probably theabsence of any quantitative taste or odour analysis of samples from thenetwork by their laboratory. The company took 269 samples in thecourse of 28 investigations covering 111 consumer contacts (theremaining were routine scheduled samples) between 4 January and 15February inclusive (before the contaminants were known). Over thewhole event up to 24 March 2010 Thames Water collected a total of 414samples. In the equivalent event period (1 February to 19 March 2010)Essex and Suffolk Water took 87 samples at customer taps of which 19were in response to consumer contacts and 14 were resamples and 12had a description of the odour. The company also took 200investigational samples at Chigwell WTW of which 46 exhibited adefined odour and in some cases a taste also.4.8. Both companies took a similar percentage of samples in response tocontacts and as part of investigations (approximately 18%). The contactrate for Thames Water was three times higher at 0.6 contacts per 1,000population (average), for Essex and Suffolk the contact rate was 0.2 per12

1,000. Therefore if all other factors had been equal the likelihood ofdetection of the odour by Thames Water was higher. The Inspectoratetherefore considers there would be of merit in Thames Water carryingout a detailed review in order to understand why they were unsuccessfuland how their diagnostic processes can be improved upon for the future.4.9. The similarities with the River Severn (Wem) pollution incident in April1994 are numerous. In both events the odour was noticed at the sewageworks but not considered significant, and no odour was detected in theriver or at the water treatment works until consumers began reporting theproblem. None of the staff either in 1994 or 2010 detected an odour onsite from the raw water intake through to treated water at taps. In 2010Thames Water did not detect an odour in any sample throughout theevent, (other than the anecdotal report from staff at the Woodford Depoton 10 February 2010). The identification of the causative compoundstook considerable time; two weeks in the case of the River Severn(Wem) pollution incident and a week in 2010 if the start point was takenas 11 February, although considerably longer otherwise. However, a keydifference is that in 1994 the National Rivers Authority (now theEnvironment Agency) took 16 hours to pin the source to the Wemsewage treatment works, 120 km upstream from Barbourne WTW. Thereason why the NRA were successful was for twofold; the personnelknew the odour they were “looking for” having been provided with adescription and secondly and most importantly they used a specificsampling and testing procedure where water was taken into a samplingvessel, shaken and then vapour was sniffed rather than water decantedfrom the bottle for assessment.4.10. In the case of Essex and Suffolk Water, the company recognised theneed to adapt their taste and odour protocol which then started toprovide diagnostic results Unfortunately the company had failed initiallyto carry out the test at 25 0 C, (a requirement of Regulation 16(5)(d)),which meant they missed the early warnings. By chance, the companyhad retained a sample overnight and repeated the test the next day. Thisled to the first detection from a sample taken at the works on the 9February.4.11. Essex and Suffolk Water were using a smell bell which is a device thatsprays water into an upside-down glass bell with an aperture at the top.(An example of which can be seen in the Standing Committee ofAnalysts publication 4 ). This device enhances odours by creating avapour. The water should be heated to 25 0 C as this further enhancesany organic odours. In 1994, Seven Trent Water reported that the odourcaused by these organic contaminants could not be detected at anytemperature below 25 0 C 3 . Essex and Suffolk were running their smellbellsat 20 0 C. Thames Water was not using smell bells contrary to thethird recommendation of the report on the Wem incident to use smellbells. Thames Water and none of its staff at the treatment works ortaking samples were trained or screened for their ability to detect odours.The lack of training applied equally to Essex and Suffolk Water staff. Thefourth recommendation of the report on the Wem incicent recommendedthat staff involved in this work should be trained and tested for theirsuitability. The Inspectorate notes that Thames Water did not appear to13

adapt to the developing situation and in particular did not appreciate thatsome odours develop over time requiring changes to the routine testingprotocol.4.12. Many tastes and odours are caused by decomposition of plants, algaeand fungi and as both companies were working on this assumption as tocause, Thames Water reinstated the out of service ozone plant atCoppermills WTW on 10 February and Essex and Suffolk Water‟s WaterQuality Team requested that the ozone plant at Chigwell WTW bereturned to service on 11 February and following re-commissioning, itwas put into service on 12 February. Additionally, Thames Watercommenced algal analysis on 11 February. The decision by bothcompanies to reinstate the ozone plant was reasonable in the absenceof other information. Such treatment is beneficial in removing tastes andodours including geosmin and methyl isoborneol and together withtreatment with granular activated carbon removes a wide range oforganic compounds from water. However for the particular contaminantsin this event the benefit of ozone is small.4.13. In summary and because of the failure to either interpret, understand oridentify initially the cause by Thames Water, there was a significantdelay in taking effective remedial action for about a fortnight in earlyFebruary. Some of the early decisions, whilst sensible in the context ofthe dearth of understanding and knowledge at the time, were adistraction, particularly the reinstatement of the ozone plant, and thenincreasing the dose twice which imposed additional operational effortand risks to water treatment to be managed. Decisions were based ongeneralised assumptions about raw water seasonal variability, there wasnot a focus on the specific taste and odour event information being givenby consumers. This was in stark contrast to the Wem event where thecompany used taste and odour information to guide operationaldecisions. Accordingly the Inspectorate notes with concern that therewas a failure of application of risk assessment principles and this revealsweaknesses in the water safety plan methodology as applied.14

5. Identification of the Contaminants5.1. On 16 February Thames Water obtained analytical confirmation that thespecific cause of the taste and odour was due to the presence of the twochemicals 2-ethyl-5,5-dimethyl-1,3-dioxane (2-EDD) and 2-ethyl-4-methyl-1,3-dioxolane (2-EMD). The information that led to this discoverywas acquired through mutual support between laboratories and scientificexperts who recognised the event as having features in common withthe 1994 River Severn (Wem) pollution incident of April 1994 3 . This wasa key observation which was made on 15 February 2010. Thisreconfirms the value inherent in industry-wide collaboration to managesituations outside the everyday. The Inspectorate commends the use ofmutual support but notes that its value was not acknowledged in theThames Water report. The company needs to reflect again and in moredepth on the benefits to be derived from promoting more activeparticipation by its staff in industry wide networks and forums to maintainknowledge and to keep itself up-to-date with new knowledge. Allcompanies would benefit from similar reflections about investing in andsharing knowledge as a matter of course.5.2. The identification of contaminants capable of causing taste and odourproblems from raw water pollution incidents poses special monitoringdifficulties and this was a particular feature of this event. The compoundsare highly soluble polar compounds, low in concentration and they posedifficulties in extraction, concentration and detection. It is thereforeunsurprising that the company did not detect these compounds initiallyon GC/MS analysis.5.3. There have been several documented reports of drinking watercontamination with 2-EDD and 2-EMD. The first was reported in thejournal Analytical Chemistry in August 1993 6 . This occurred atNeshaminy WTW owned by the Philadelphia Suburban Water Companyin Middletown Township. In January 1992 a commercial waste facilityaccepted wastewater from a resin coatings manufacturer. This directlyresulted in an objectionable taste and odour experienced by consumersfor two weeks. It took the company approximately 20 days to determinethe cause which then led to the identification of the two as then unknowncompounds.5.4. The second incident was the Worcester or Wem incident in April 1994 3where on the 11 April 1994 organic chemicals were discharged into asewer by Vitalscheme Limited (who specialised in the recovery ofsolvents used in the production of resins) passed through the Wemsewage works and then on into two small tributaries feeding the RiverSevern, most likely on 12 April 1994. Three days later on 15 April,consumers began to complain of a taste and odour. The identification ofthe organic chemicals took two weeks. The report into the incident,published in August 1994 considered this to be an excessive time andthat much time would have been saved had a paper describing thePhiladelphia incident come to light earlier.5.5. From October 1993 the Tordera aquifer located in Barcelona 7 wascontaminated by a factory producing saturated and unsaturated15

polyester resins and sited in the upper course of the river. This causedcontamination of a number of boreholes used for drinking water supplypurposes with subsequent taste and odours. As a consequence of thisstudy which looked at 48 groundwater supplies over 1994 and 1995,complaints from consumers of Barcelona‟s tap water were noted inNovember 1994 and January–February 1995. This led to the discoveryof the contamination of the Llobregat River 15Km from Barcelona‟sWTW by a factory specialising in the elimination of industrial residues.5.6. In September 2003 in South America 8 a major works to several millionconsumers had to be shut down due to many consumer complaints. Inthis particular instance there were a number of contaminating chemicalswhich prevented easy identification of the cause of a characteristic oliveodour. After six weeks both chemicals were isolated and it was thisinformation that led to the identification of the source which was apolyester resin manufacturing plant.5.7. In all these cases the identification of the causative chemicals wasdifficult. In this case Thames Water identification of the chemicalsoccurred on 16 February 2010, five days after the company was alertedto the fact that the problem was one of raw water by Essex and SuffolkWater on 11 February 2010. It was a day later that Thames Waterappeared to take remedial action because on 12 February the River Leeintake was closed, the ozone dose was turned up and the laboratoryrequested assistance. The 1994 Wem incident was resolved within thistimescale therefore it was time taken by Thames Water to recognise araw water event, rather than the time taken to identify the chemicals, thatwas the issue on this occasion. Notably, the identification of thecausative chemicals in the Wem incident was a secondary considerationbecause the event was solved by the most practical of ways, sourcetracking using the human nose. In hindsight the criticism of thelaboratory in the report of the Wem incidentt 3 was overly harsh given thatonly one previous event world wide had been recorded at the time 6 . TheInspectorate‟s own report on the Wem incident (Number 36/2/1) did notendorse this criticism of analytical science and this is the case in thisevent also because by 15 February the laboratory mutual networksuggested the answer and on 16 February the laboratory was able toidentify 2-EDD and 2-EMD. Indeed the Inspectorate notes this was thefastest identification of these compounds in any of the recorded andknown events worldwide. The industry is reminded of the value to bederived by collective investment in the knowledge and informationsystems and expertise to be called upon promptly.5.8. For the purposes of industry learning and understanding the massspectra of both organic compounds together with a brief outline ofmethodology is provided in Appendix VI.16

6. Source of the Contaminants6.1. Rye Meads STW is located about 10 miles upstream of the intake to theLee Valley reservoirs. Sewage arrives at the works via three sewerscarrying sewage sludge, waste from inter-site tankers and a dedicatedcommercial waste stream. The treatment processes comprise ofpreliminary treatment, to remove grit and screenings, followed byprimary sedimentation and activated sludge treatment in aeration tanks.After settlement of the activated sludge, the effluent passes through alagoon system to provide further effluent polishing. The effluent ismonitored for suspended solids, BOD, ammonia and phosphate. Sludgefrom the primary and secondary settlement stages is thickened andblended and then passes through primary and secondary digesters afterwhich it is dewatered by presses. The solids from this process arerecycled to land and the water fraction returns to the head of the worksand passes into the effluent line for processing through the works.6.2. The commercial sewage stream, managed by a large commercial wastecompany, enters at the blending point prior to primary and secondarydigestion. The waste company was responsible for the receipt of thewaste, and sampling and blending of their tankers. Control of this is bychemical oxygen demand and volume up to a limit of 270m 3 /day. Sitequality control for their own waste stream was carried out within theterms of the contract set up in April 2004 which required only waste thatis biodegradable and contained specific exclusions such apharmaceuticals, petroleum and the Environment Agency list of 88substances of concern. The waste company assessed their own clients‟waste however it was the responsibility of Thames Water to ensure thatthe waste did not affect the treatment process, the effluent quality, theenvironment and specifically the quality of the raw water abstracted forwater supply purposes. This was assessed through analysis twice aweek for ammonia, metals, solids, COD and oils and four times a yearusing GC/MS for organics. The contract was managed by a SeniorConsultant for Waste Water employed by Thames Water. The sitemanager and operational staff at Rye Meads had no control over thecommercial waste company activities nor did they have knowledge of theindividual clients of the commercial waste provider because thisinformation was classed as confidential.6.3. Rye Meads STW is located within the Chingford South and the EnfieldLock Intake community and was deemed to be high risk due to itsproximity to the intake. The risk assessment required that the companyidentify the highest risk potential contaminants through consideration oflicence and discussion with the Environment Agency as necessary. Thecompany had committed to monitor these sufficiently to validate the riskassessment under normal conditions.6.4. The commercial waste company was identified as part of the drinkingwater safety plan, (DWSP), but not specifically in the risk assessment,although the consequential risk score was said to be in the highestcategory of risk available. Thames Water within the risk assessmentconsidered that this was an acceptable risk through mitigation by audit,17

eview and implementation processes. However, the agreement with thecommercial waste company which lists the Environment Agencychemicals of concern did not list either 2-EMD or 2-EDD. The riskassessment produced by Thames Water with assistance from anexternal consultancy did not include identification of 2-EMD or 2-EDDbecause the person who was responsible for the risk assessment did notknow of the existence or significance of these two chemicals.6.5. Both water companies‟ risk assessments identified contamination of theRiver Lee by sewage effluent as a separate hazard and listed the risksas BOD, Ammonia, Solids, and Dangerous/Priority substances. Bothcompanies considered these risks to be mitigated and thereforeacceptable. The only unacceptable risks assessed by Thames Water forthe Chigwell community intake, (the raw water supply to Essex andSuffolk Water), was operator competency with poor visibility of staffcompetency and inadequate system for training record updates. Theonly unacceptable risk at Coppermills WTW was the risk associated withpower loss and the lack of emergency power back-up for the site. In2009 the company included a scheme to address this risk in its businessplan for the Ofwat price review process.6.6. The Rye Meads STW manager and Process Control Engineers, (PCE‟s)were not involved in, or aware of, the water safety plan process andmethodology. In practice their only involvement with staff at CoppermillsWTW was to highlight a risk when the ammonia concentration in effluentincreased when they would contact staff at Coppermills WTW to alertthem of the need to close the intakes.6.7. After identification of the contaminants, Thames Water sampled the rawwater reservoirs on 16 February 2010. Within the Lee Valley storagereservoirs on 16 February the highest concentration of 2-EDD was0.352µg/l and for 2-EMD this was 0.192µg/l (semi-quantitative analysisonly) in the William Girling reservoir.6.8. Thames Water‟s report states that on 16 February 2010, on-siteinspections were conducted at all of its 26 Sewage Treatment Works inthe catchment and this identified an unusual odour associated with aprocess stream treating commercial waste at Rye Meads sewagetreatment works, (STW). In statements given to the Inspectorate, neitherthe Rye Meads Team Manager nor the Process Control Engineers,mentioned this. One knew about the drinking water taste and odourproblem in North London via an internal e-mail around the 12 February2010 and another became aware by word of mouth on 21 February 2010and then subsequently through a call from the catchment Manager on 22February 2010. The Manager was on leave at the time.6.9. On 18 February 2010 analysis of a Rye Meads STW final effluent from aroutine retained sample from 12 February detected 2-EDD at 2.52µg/land 2-EMD at 1.761µg/l. Samples by the Environment Agency on 24February detected both 2-EDD and 2-EMD downstream of Rye Meadsbut not upstream. These findings proved beyond reasonable doubt thatthe cause of the taste and odour issue was the presence of low levels of2-EDD and 2-EMD in the River Lee was treated sewage effluent fromRye Meads STW.18

6.10. Retrospective analysis of a four week combined Rye Meads samplefrom January from the dedicated waste stream gave values in this wastestream of 8.1mg/l 2-EDD and 2.8mg/l 2-EMD. There was no historicmaterial for December 2009 so it was not possible to determine whenthe contamination first started. Thames Water was also able to samplethe head of the works before and after the liquid return from thededicated waste stream and this did not identify any of the compoundsbefore the return and showed levels of 1.36µg/l and 0.36µg/l for 2-EDDand 2-EMD after liquid return This conclusively proved that the sourcewas the commercial waste. The contamination persisted into Februarywith retrospective samples of the waste stream from 13 Februarymeasuring 5.6mg/l and 1.3mg/l for 2-EDD and 2-EMD respectively.6.11. Thames Water collected samples from traders whose waste had beenhandled and discharged at Rye Meads STW via the commercial wastecompany However, this testing did not yield information to identify thetrade source. Subsequently a review of other traders whose waste ismanaged by the commercial waste company was conducted andsamples examined. This exercise identified one company where thetarget compounds were identified at high concentrations (2000µg/l and2350µg/l respectively). Investigations identified this trader as a resinmanufacturer. This source is common to all historical incidents involving2-EDD and 2-EMD. It is however also worth pointing out that the mixingof wastes containing aldehyde and glycol may also have the sameoutcome.6.12. The sewage treatment including blending the sewage had the effect ofreducing the contaminant level (hundred to one thousand fold) comparedto that found in the final effluent at the outfall.6.13. 2-EDD is the most odorous and distinct of the two compounds asdemonstrated by the historical incidents. The report on the 1994 Wemincident 3 made a series of recommendations but the firstrecommendation states that “An observation on any unusual smelldetected in an effluent should be a standard requirement in routine tests,at sewage treatment works.” The second recommendation required amethod to assess the odour of effluents to be developed, taking intoaccount health and safety of the staff.6.14. At Rye Meads STW in 2010, there was no formal odour monitoring otherthan the odour management plan for external public complaints whichare followed up by the Team Manager. In the most part he wassuccessful in resolving complaints, but not always, which he thenescalated within the company. However, any subsequent action and itseffectiveness when odours were found is questionable. This wasparticularly the case when any odour was not associated with acomplaint but only noted on site. It is notable in statements to theInspectorate that the staff were disinclined to report such matters inrecent years because no action was taken, most notably when there wasa previous petrol like odour. This situation represents a missedopportunity by Thames Water because relevant odours had beennoticed and investigation of them may have identified issues with thecommercial waste stream.19

6.15. One of the two PCE‟s had a call from the catchment Manager on 22February 2010 to advise him of the taste and odour event and to arrangefor the shutting down of dewatering, (liquor return) from the commercialwaste stream because the taste and odour had been traced back to thesludge stream. The return was stored on site and subsequently tankeredaway for disposal in another way with the agreement of the EA. TheTeam Manager was only informed of this decision by the CatchmentManager on 23 February and the process Manager was on leave at thetime. The process Manager understood that the commercial wastecompany had been stopped from putting any further waste into RyeMeads, however, the letter to this effect from Thames Water was notsent until 4 March 2010.6.16. It was therefore at least 11 days before the source of the contaminationwas stopped and probably the contamination was occurring for a periodof seven weeks, assuming that all four weeks of the combined samplecontributed to the contamination, or longer using the reservoir turnoverinformation described in the next section. This duration comparesunfavourably with that of the 1994 Wem incident20

7. Raw Water Management7.1. In the absence of information suggesting that Rye Meads STW wascontaminating the River Lee, Thames Water continued to abstract waterinto the raw water reservoirs up until 12 February. This could have beencontinuing for at least six weeks based upon the combined effluentsample, or longer if based on consideration of the turnover rate of thereservoirs.7.2. The River Lee at the Rye Meads effluent channel is on average between0.22 to 0.65 meters deep and the flows at Fields Weir was on average8.3 cumecs (range 3.8 to 37.6) between January and March 2010. Incomparison the River Thames flow at Walton Bridge in West London ison average 53 cumecs, (max 135). The River Lee is therefore arelatively small river. The discharges from sewage treatment worksaugment flows and support abstractions further downstream. Out of the178Ml/d consented discharge into the catchment, 101Ml/d (57%) is fromsewage treatment works. In some stretches, during long dry periods,discharges from STWs make up the only flow in the river 11 . Rye MeadsSTW together with Deephams STW are the source of the mostsignificant sewage effluent discharge into the River Lee, and this forms12 aspart of The London Catchment Abstraction Management Strategypublished by the Environment Agency7.3. Based upon the analytical data, the River Lee dilution capacity betweenthe outfall and the intake is from as little as 7 fold to 269 fold for 2-EDDand 3 fold to 65 fold for 2-EMD. The variance is likely to be due tochanges in flow volume of the river and volume of outfall, as well asanalytical and sampling uncertainty during the period. The distancebetween outfall and intake is around 10 miles so comparison has beenmade of data from the same days. The river being primarily composed ofdischarge clearly contributes only a small dilution effect of thecontaminants from Rye Meads. As this event occurred inJanuary/February the river would be expected to offer probably the bestdilution possible, when compared with the summer when river levels aresubstantially lower. In one risk assessment, the contribution to the riverof effluent in the summer from Rye Meads is recorded as being as highas 90% by volume. This is clearly an important factor to be taken intoaccount in any drinking water risk assessment.7.4. The River Lee on 12 February at the intake of the William GirlingReservoir contained 0.037µg/l 2-EMD and 0.013µg/l 2-EDD. The daybefore, the levels in William Girling were 0.154µg/l and 0.077µg/l and inKing George V 0.042µg/l and 0.025µg/l for 2-EMD and 2-EDDrespectively. In effect the concentration of contamination in thereservoirs was higher than in the river at the intake on that date. A higherconcentration of contaminants in the reservoirs would imply that thewater was either completely replaced from the inlet with a concentrationthat was previously higher, or if not completely replaced, then theconcentration from the intake was significantly higher prior to 12February 2010.7.5. This is a very important observation because the nominal residence timecalculated by volume for the William Girling reservoir is on average 11621

days and at maximum flow is 89 days (divided by flow). This calculationimplies all the water is replaced, but assuming this does not happen inpractice due to factors of mixing/flow, then a calculation of 33 – 43 dayshas been assumed. This means the movement of a chemical from riverto the intake would take between a week to ten days. Between 11 and24 February 2010, the concentration of 2-EMD in the William Girlingreservoir was twice that at the intake and four times for 2-EDD. The levelof 2-EMD in William Girling was 10 -15 times above the odour thresholdand in the King George V reservoir, 4 times higher than the threshold.This circumstance could not be explained by the ten day direct path asmentioned above. This points towards the importance of theembankment works on the King George V reservoir which meant theintake was closed between 26 December 2009 and 14 January for allbut three days and Coppermills WTW remained in constant use at 8%draw down. On the week beginning 8 January 2010 the amount takenfrom King George V reservoir into Coppermills WTW increased by nearly10% to 17%, (increase between 20 Nov and 5 February 2010 was a6.8% to a 20% contribution), and the demand from the Thames LeeTunnel which takes water from the River Thames fell by 6% to 14%,(decrease between 20 Nov and 5 February 2010 was 21% maximum toan 11% contribution) (Appendix II). This River Thames water was freefrom contaminants. The first spike in consumer complaints occurredshortly after this time and therefore the reservoir must have beencontaminated on a date before 26 December 2009. It is likely that thisroutine change in the raw water management regime (mix) caused thecontaminant concentration which was already likely to be within thereservoir to increase to above the taste and odour threshold enablingsensitive consumers to discern its presence.7.6. The closure of the intake (downstream of Rye Meads STW) on 12February, whilst an important action of remediation, was too late to bepreventative. However, it is important to note that had Thames Waterimplemented learning from the 1994 Wem incident (thirdrecommendation) by implementing improved odour tests then the earlywarning information would have been available to enable the intake tohave been closed earlier. The recommendation suggested the use ofsmell bells at works and whilst Thames Water had initially followed thisadvice, they were removed by the company quoting health and safetyreasons. Had this decision not been made then the 2010 event mayhave been prevented.7.7. Coppermills WTW can be directly supplied through the CoppermillsStream and the Spine Tunnel from William Girling and King George Vreservoirs and/or by Walthamstow reservoir and East Warwick reservoirwhich are in turn fed by High Maynard reservoir. Analytical data showedthat all routes feeding Coppermills WTW contained both contaminants insome proportion and this is consistent with all these sources receivingRiver Lee water abstracted below Rye Meads STW in some proportionbefore changes to the raw water supply arrangement. As an example,0.455µg/l of 2-EMD was measured in the Coppermills Stream on 21February 2010 ten days after the intake closure. Changes in raw waterconfiguration (with the proportion from the Thames Lee Tunnel rising to22

36%) shows the influence of River Thames water upon the whole event.High Maynard Reservoir, Coppermills Stream and the Spine Tunnel arefed by the Thames Lee Tunnel. It is unsurprising that the consumercomplaints continued to rise to a peak on 16 and 17 February before asubsequent decline as the water supply system was cleared ofcontaminants over five days in line with the increased flow of raw waterfrom the Thames Lee Tunnel7.8. When the company first became aware of the raw water problem theyclosed all the intakes from not only the River Lee but also from NewRiver and contacted the EA. In hindsight the closure of the intake fromthe New River proved to be unhelpful because this river divides from theRiver Lee upstream of Rye Meads and was therefore free ofcontamination. Seventy percent of the raw water supply for the KingGeorge V reservoir came from this source and so by closing this intake,an opportunity was missed to maintain a contaminant free intake for thereservoir which would have diluted the existing contamination andprovided a better quality supply to both Chigwell WTW and ChingfordWTW. A decision supported by reliable odour monitoring would haveaided the decision about remedial raw water management.7.9. The raw supply to both Chingford WTW and Chigwell WTW could not bechanged away from either the William Girling or the King George Vreservoirs. The supply to Chingford WTW remained a blend ofapproximately 70:30 respectively and for Chigwell WTW the bulk supplywas left as being solely from William Girling reservoir. The reason forThames Water supplying Essex and Suffolk Water entirely from theWilliam Girling reservoir was because the levels of water in King GeorgeV storage reservoir were already low due to maintenance activities onthe embankment. The embankment works therefore proved to be acontributory factor to the event duration as they had the effect ofremoving an alternative supply to Chigwell WTW. As the William Girlingreservoir receives its entire supply from the River Lee it consequentiallyhad a higher concentration of the contaminants and this was confirmedlater by analysis. However, had the source of the contaminants beenknown, this fact would have been obvious and Thames Water couldhave, and should have, changed the bulk supply to Essex and SuffolkWater to being entirely from King George V reservoir.7.10. The Inspectorate is critical that there was no alternative arrangement forwater which is exported to another company and also that there was nobulk supply agreement with quality monitoring to prevent this situationfrom arising contractually. At the time an unusual odour was noted in asample taken from William Girling so the company were aware that thebulk supply was unsuitable therefore the decision was for reasons ofquantity (level in the King George V reservoir) rather than quality.7.11. The concentration of both contaminants on 11 February in King GeorgeV reservoir was approximately one-third of that in William Girling, theproportional difference to the volume of intake from the New River to thatof the River Lee. Thames Water changed the bulk supply for Essex andSuffolk and incrementally changed the supply for Chingford WTW to70% from the King George V Reservoir on 17 February and reopened23

the New River intake only after it was aware, through analysis, of therelative contaminant levels.24

8. Final Water Management8.1. In this event Coppermills WTW became the main focus for ThamesWater because the area supplied by this works was receiving twice therate of consumer contacts compared with the area supplied by ChingfordWTW. Chigwell WTW was managed by Essex and Suffolk Water but itreceived a bulk supply of water from Thames Water and shouldtherefore have been an equal focus of attention and information.8.2. There are two treatment processes in use at Coppermills WTW designedto control pesticides, but also capable of dealing with taste and odoursand these are ozone and granular activated carbon (GAC) installedwithin slow sand filters. The process diagrams can be seen in AppendixIII.8.3. In October 2009, Thames Water refurbished the plant installing newdielectrics. However, due to operating problems, the ozone plant was notoperating between October 2009 and February 2010. Ozone dosing atCoppermills WTW was re-instated on 10 February in response to theevent, but as discussed earlier, this was because the company wasconsidering the cause at that time might be algal.8.4. Following re-introduction and assessment, ozone concentrations wereincrementally increased during the event from 1.5mg/l to 3.75mg/l inresponse to emerging information.8.5. The company carried out a number of checks at the treatment worksincluding filtration, GAC and disinfection, as well as organising siteaudits by its own scientists. The company found no treatmentdeficiencies or network activities that might have been contributory to theevent.8.6. Essex and Suffolk Water also examined slow sand filters and networkactivities and found nothing abnormal other than the raw water resultingin a contact to Thames Water.8.7. Thames Water sampled at a number of locations includingretrospectively on retained samples and both compounds were identifiedin treated water from Coppermills WTW. Sample results in CoppermillsWTW treated water contained concentrations of 2-EMD ranging from0.124µg/l to 0.156µg/l and of 2-EDD ranging from 0.016µg/l to 0.026µg/lbetween 15 and 19 February 2010. Chingford South WTW contained aresult of 2-EMD of 0.037µg/l and 2-EDD of 0.0005µg/l on 18 February.Chigwell WTW contained a result of 2-EMD of 0.018µg/l and 2-EDD of0.0005µg/l on 25 February as the single result. Sampling at the worksand in distribution continued and showed concentrations declining overtime following the remedial action in relation to intakes described above.8.8. The quantity and quality of the data, together with the inherent variabilityof the analytical methodology and sampling and also the lowconcentrations of 2-EMD and 2-EDD observed, all make extrapolationof the effectiveness of their removal by the water treatment processesproblematical. However, there would appear to have been minimalremoval at Coppermills WTW and only some removal at Chingford andChigwell WTW. These observations confirm that no particular treatment25

process is favourable for removal of these chemicals. The threetreatment works included between them, all conventional processes,ozone, coagulation, and both rapid gravity filters and slow sand filterswith GAC. This observation is consistent with the scientific literature8.9. Thames Water took samples at four treated water service reservoirs andthe earliest of these samples was from 22 February at Woodford ForestService Reservoir. The maximum value measured was 0.146µg/l 2-EMDand 0.015µg/l 2-EDD on this day. Thereafter the values reduced in allreservoirs. The Inspectorate has noted that Thames Water did notsample at Stewardstone service reservoir until 11 March 2010 which isnot consistent with a timely event response.8.10. Sampling was carried out at consumer taps between 6 February and 16March. The maximum values found were 0.186µg/l 2-EMD and 0.022µg/l 2-EDD on 18/19 February 2010. This was one week after the RiverLee intake closure and coincides with the peak number of consumercontacts. After this date the levels reduced although they remainedabove the odour threshold of 0.1µg/l for 2-EMD until 9 March and for 2-EDD on 25 February 2010. Again this coincides with the decline inconsumer contacts.8.11. There was no sampling undertaken by Essex and Suffolk for thesecompounds at consumer taps.26

9. Health Protection Agency Liaison9.1. Thames Water held an initial meeting with the local Health ProtectionUnit (HPU) of the Health Protection Agency on 9 February 2010 andthen a follow up meeting took place on 10 February 2010 Notes of thesemeetings were not provided to the Inspectorate by Thames Water andthe company is reminded that all information in respect of an event mustbe included in the company‟s final event reports. The meetings wereheld at a point in time before Thames Water had determined the causeas raw water matter, (Essex and Suffolk where still being supplied withthe contaminated bulk supply). At this stage the company advised therewas a problem with the chloramination process and HPA advice wasgiven on this basis. It was also reported at this time that about 10% ofconsumers were reporting symptoms of diarrhoea, headache and skinrash. The HPA advised that there was no increased level of potentiallywater related illness in the population of the affected area; however, theDirector of the North East/North Central London HPU was aware of alarge outbreak of norovirus, (a group of single-stranded RNA viruses thatcause acute gastro-intestinal illness lasting 2-3 days, self limiting andoften transmitted by hand to hand contact, surfaces and the faecal-oralroute), affecting the catchment of Whipps Cross Hospital. It wasconsidered that the norovirus outbreak was not related to water and itwas noted that disinfection was operating effectively for virus removal.9.2. Essex and Suffolk Water notified the HPU on 11 February and soughtindependent medical advice on 13 February 2010 with their contractedadvisor who also had also attended the HPU meeting on 10 February2010. Their advice was to monitor the situation in close liaison with theHealth Protection Agency.9.3. Thames Water became aware of the two chemicals, namely 2-EMD and2-EDD on 15 February through the mutual support network and soughttoxicological advice on that day. The company identified both chemicalsin samples on 16 February 2010 and due to this new information called ameeting with the HPA. The meeting held over two days ( 17 and 18February) included Essex and Suffolk Water, the Chemical Hazard andPoisons Division, (CHaPD), Health Protection Agency (HPA) and on the18 February the Inspectorate was in attendance.. CHaPD providedwritten advice on 18 February 2010 which stated that assuming thelaboratory analysis was suitably quality controlled then the reportedconcentrations are several orders of magnitude lower than the levelsexpected to cause health effects in humans.9.4. Results from Coppermills WTW s confirmed concentrations of 2-EDD of0.024µg/l and 2-EMD of 0.143µg/l in drinking water, above the 0.01µg/ltaste and odour threshold., therefore it was concluded that these levelsmay lead to consumer taste and odour complaints.9.5. Following the Wem incident there were two health studies carried outand published in 1996 and 1998 respectively. The first study investigatedwhether exposure to tap water contaminated with 2-EMD and 2-EDDwas associated with an increase in self reported symptoms betweenthose who noticed an unusual taste and odour and those who did not.27

The second study looked at bias in self reported symptoms to anemergency helpline. In the first study by S. E. Fowle 10 et al, 62% ofsubjects, (867 of 1398) in the study group had noticed the taste andodour. Among those who had not noticed any unusual taste and odour,there was no association between drinking the water and reportingindividual symptoms, only between increasing dose and nausea. Amongthose who had noticed the unusual taste and odour, there was both anassociation and a dose response between the water and the reporting ofdiarrhoea, nausea, headache, stomach pains, skin irritation and itchyeyes. In the second study by David L. Fone 9 et al, 106 complainants toan emergency helpline were sent a questionnaire asking aboutsymptoms following consumption of tap water during the incident in1994. 89 complainants responded of which only 45 were in fact exposedto the contaminated water. The study found no difference in rates of selfreporting of symptoms between those exposed and those who were not.This suggests that the „worried well‟ used the helpline regardless ofwhether they were exposed. The findings from these studies have beentaken into account when assessing the response of consumers duringthe 2010 event.28

10. Consumer Perceptions10.1. From 3 January to 21 March, the total number of consumer contacts toThames Water during the event was 1114 and 104 of these mentionedillness (total population of affected area was 2 million). After excludingrepeat contacts, 848 of the contacts were unique and 12% of thesereferred to illness. In the area supplied by Essex and Suffolk Water thecontacts occurred between 3 February and 11 March and there were 97in total, of which 7 (7%) mentioned illness. It is possible that thedifference in reporting rates reflects the higher concentrations measuredin water supplied from Coppermills WTW compared to Chigwell WTW asthis would be consistent with the above mentioned published findings.The peak consumer response in terms of number of unique contactsoccurred between 15 and 19 February 2010.10.2. Essex and Suffolk Water‟s report describes how their water qualitydepartment liaised with their communications teams and call centre toensure that the information being given out to consumers was up to dateand accurate. Written guidance was issued to call centre staff on 15February. On 19 February 2010 Thames Water updated their customercentre with public health information including the names of thechemicals identified, the source (River Lee) and the advice given by theHealth Protection Agency. Further updates occurred over subsequentdays.10.3. It is not uncommon for the first indication of a problem to be contactsfrom consumers. In the 1994 Wem incident the first customer contactwas recorded at 07:50 hours, multiple contacts (20) were received by09.40 hours and within 80 minutes the treatment works had been shutdown. However the situation in the 2010 event was not directlycomparable because the scale of the works in the Wem incident wasrelatively small (served 30,000 consumers). Looking at the Essex andSuffolk Water situation, the population was an order of magnitudegreater (460,701 consumers). The first consumer contact to thecompany was received on 3 February 2010 but it was only three dayslater that it became evident that there were multiple contacts (on 6February 2010 there were three contacts) and by the time Essex andSuffolk Water called Thames Water there had been 31 contactsrecorded. After taking into account the larger scale and complexity(multiple zones) the response of consumers in the Essex and Sufffolksupply area is comparable to that recorded for the 1994 Wem incident.In the case of consumers in the Thames Water supply area there were166 contacts received upto and including 8 February. Although theabsolute number is four times greater than experienced by Essex andSuffolk Water, the contacts came from 42 different zones compared tojust eight zones for Essex and Suffolk Water. The question arisestherefore as to whether it requires five times the number of consumercontacts to be received before an event can be recognised across anarea which comprises five times as many zones. The findings from the2010 event would imply that the scale and complexity of a supply area isa rate limiting factor in respect of speed of event recognition. A29

confounding factor in establishing retrospectively the exact start of theevent in the 2010 event was the Norovirus outbreak in the communitycentred on Whipps Cross and also the retained effluent sample whichwas a pooled sample representing a four week period of time. Using justthe Essex and Suffolk Water consumer data would identify the start ofthe event as the beginning of February and this is not inconsistent withthe Thames Water consumer data which showed a noticeable increasein reporting towards the end of January into the beginning of February.This suggests that it was only in late January, rather than December,when the chemicals reached or exceeded the threshold taste and odourconcentration in tap water sufficient to evoke a consumer response. Insummary the operational consequences of the change in the raw watermanagement regime to accommodate maintenance of the King GeorgeV reservoir appear to have been critical to raising the concentration ofcontaminants in the raw water to a level sufficient to impact on tap waterquality and hence consumers.10.4. The 2010 event reinforces the value of tracking consumer responsesonce an event has been recognised and facts illustrate that ThamesWater did not perform well in this regard. Essex and Suffolk Watertracked six consumers by call back on 12 March, (and attempted butfailed to contact another ten consumers) and this strategy provedsuccessful in determining that residual post event concerns had beenresolved. Thames Water tracked 11 consumers (a 1% sample comparedto the 6% sample of Essex and Suffolk Water). Overall thereforeThames Water did not sufficiently determine that consumer concernswere resolved and this lack of follow through is reflected in thecompany’s report which did not consider such matters. However thecompany did write to all consumers who had contacted them during theevent and this letter was sent on 26 March 2010.10.5. There were 16 consumers from the area supplied by Thames Water whomade direct contact with the Inspectorate (2% of unique contacts inresponse to the taste and odour event) and the equivalent numbers (3)were received from Essex and Suffolk Water‟s consumers. The firstcontact from a consumer in the Thames Water area was received by theInspectorate on the 10 February 2010 and the last on 5 March 2010. Thefirst contact to the Inspectorate from a consumer of Essex and SuffolkWater was on 18 February 2010 and the last was some months later on28 June 2010. The peak for consumer contacts to the Inspectorateoccurred between 19 February and 23 February (typically 3 or 4 a day).The number reporting illness complaints represented 12.5% of the total(exactly in proportion to the illness reporting to the companies).10.6. The first contact to the Inspectorate, who was a customer of ThamesWater, reported a strange taste and odour in tap water to the companyfrom mid-January onwards before contacting the Inspectorate. Theconsumer expressed a lack of trust in the company. It was said thecompany was very reluctant to admit initially that there was a problem.This was a recurrent theme in the subsequent 15 contacts to theInspectorate; the time taken to acknowledge there was a problem wascrucial to consumer perception, as was the lack of any statement froman independent health advisory body about the risk posed by the30

chemicals found. The relatively low key media coverage caused someconsumers to believe that there had been “a cover up”. Theseperceptions are a direct reflection of the uncertainty as to cause withinThames Water and this can be tracked exactly by the experience of thefirst consumer to contact the Inspectorate. He was told firstly that therewas a problem with the chloramination process and then by 18 Februaryhe was being told about the presence of the odoriferous chemicalcontaminants. His experience independently verifies the informationbeing given out by the company at various points in time but it alsoillustrates the fragility of consumer confidence in the face of a protractedevent where information changes over time. For the majority ofconsumers their concerns dissipated when their supply returned tonormal however for a minority the event resulted in a more significanterosion of confidence in drinking water safety.10.7. The Inspectorate directly sought the views of all those consumers whohad contacted the two companies during the event using a standardevent questionnaire. This group of consumers is not a representativesample of the population potentially affected instead it comprises thoseconsumers who had been sufficiently motivated to call their watercompany in response to the event. It is of interest to note that about 4%of those who completed the questionnaire stated that they did not noticeanything different about the water supply. All had stopped using thewater for drinking, cooking or washing or a combination thereof and allheard about the event through the media. Almost two thirds noticed theproblem in February with the remainder noticing it first in January. Threequartersstated they were given advice by the company as aconsequence of making the phone call, the rest did not recall receivingany advice. Some 3% mentioned that they had been offered analternative water supply by Thames Water, compared to none of therespondents from the area served by Essex and Suffolk Water. (Fullquestionnaire data is provided in Appendix VIII)10.8. The Inspectorate visited and interviewed twelve consumers, six fromeach company. At around the time of the event all of these consumershad noticed a change to the taste and odour of their water and whilstdescriptions varied (including “onions”, “decaying fish”, “yoghurt” and inone case describing the smell as “effluent”) all found the quality of thewater at this time to be unpleasant and unacceptable. In general theconsumers all turned to using bottled water as an alternative, althoughone consumer initially continued to drink the water after passage througha filter, but this reportedly made no difference to the taste and odour.Two of the consumers reported developing skin rashes and anotherconsumer who continued to use the water (boiled and in hot drinks) latersuffered a bout of diarrhoea, for which she sought medical advice andtests. Largely the interviews verified and followed a similar pattern to theresponses given by the wider group in the questionnaires.10.9. In summary, the Inspectorate‟s own investigation obtained reliableevidence that consumers rejected the water for drinking, cooking andwashing (domestic purposes) on grounds that it tasted or smeltobjectionable and therefore the investigating inspectors concluded that31

oth water companies supplied water unfit for human consumptionduring the course of the event.32

11. Media Reporting11.1. There was significant media interest in this event primarily focussed onThames Water. Coverage included BBC Radio 4, BBC London Newsonline, BBC London News (TV), the BBC Web site, (Appendix IX), TheEvening Standard, ENDS Report and a number of local Londonnewspapers. The first of these appeared on 16 February in the form of areport on BBC London News at 18:30 hrs. The BBC followed up with anarticle on 26 February 2010. Only two newspaper articles mentionedEssex and Suffolk Water. The first report was an article in the WalthamForest Guardian on 16 February and the second article appeared in theBarking and Dagenham Post on 24 February.11.2. The first press statement by Thames Water was made on 19 February2010 following identification of the chemicals. This named the chemicalsand included the advice of the HPA as the basis for the company‟sopinion that there was no reason to believe that there was a significantrisk to public health. Subsequent press releases occurred on 24 and 26February and 02 March 2010. Essex and Suffolk Water respondeddirectly to each press statement and article.11.3. Both companies liaised with the Inspectorate in relation to media andpress interest throughout the event, although initially Thames Water‟spress office had to be prompted of the need to do so by a contact by theInspectorate on 15 February 2010.33

12. Liaison with Stakeholders12.1. The Inspectorate was notified of the event on 9 February 2010. A multidisciplinary incident management team was convened by Thames Waterbetween 3 and 10 February 2010; however the company failed toaccurately document when their incident management procedures wereevoked therefore it is questionable as to whether notification was timely,not least because 252 consumer contacts were received by thecompany between 3 January and 8 February 2010. The company hassince recognised the need to improve procedures in this area.12.2. The Environment Agency was informed by both companies on 11February 2010 via their Incident Hotline when it was known that thecause was associated with contamination of raw water. Ongoing liaisonoccurred after this date.12.3. Thames Water was contacted by Essex and Suffolk Water on 11February. No prior contact was made with Essex and Suffolk Water andboth companies have acknowledged that there was no formal agreementin place between them regarding communication or raw water bulksupply quality.12.4. Thames Water requested assistance on 11 February 2010 from theindustry‟s laboratory mutual aid group to obtain information concerningthe source of the taste and odour. The group responded with adviceabout 2-EDD and 2-EMD, spectra and standards. The Inspectoratecommends and highlights this action as being significant in managementand resolution of the event thereafter.12.5. Consumer Council for Water was notified on 10 February and therelevant Local authorities (Environmental Health Departments) werenotified between 10 and 16 February by Thames Water and on the 11February by Essex and Suffolk Water.12.6. Information on contact made with the Health Protection Agency is givenelsewhere in this report.12.7. Toxicological advice was requested by Thames Water from NCET-WRcon 15, 16 and 17 February regarding both 2-EDD and 2-EMD. Essexand Suffolk Water did not independently seek this advice which wasshared with them by Thames Water.34

13. Lessons Learnt13.1. Both companies have reviewed the event, drawn conclusions and takenactions in response to their internal findings. Thames Water intends toreview discharge agreements with trade effluent dischargers. It alsointends to review risk assessments and share this work with Essex andSuffolk Water. The Inspectorate endorses this approach, particularly thelearning point regarding sharing and transparency between companies.13.2. Both companies intend to review taste and odour methodology andmonitoring at water and sewage works as appropriate13.3. The companies intend to formalise all bulk supply agreements in relationto communications and quality.13.4. Thames Water has made changes to its handling of consumer contactsto ensure that patterns can be identified earlier and actions are taken torecord calls about quality concerns.13.5. The Inspectorate is pleased to note that findings from the event are to beshared with the wider industry at a workshop for Health and Localauthorities.13.6. Other technical actions cover analytical and treatment methodology forthe identification and removal of the causative chemicals.35

14. Contraventions of the Water Supply (Water Quality) Regulations14.1. The water supplied from Coppermills WTW and Chingford South WTWby Thames Water contravened the Water Supply (Water Quality)Regulations 2000 (as amended) by virtue of the presence of 2-EDD and2-EMD which led to the rejection of the water for drinking, cooking orwashing by 369 consumers responding to a questionnaire issued by theInspectorate. The odour threshold for both chemicals is 0.01µg/l andthere were 39 samples exceeding this level for 2-EDD and 207 samplesexceeded this threshold for 2-EMD. Out of all the laboratory tests fortaste or odour which were carried out, none were positive.14.2. There is no specific standard for these chemicals in the Water Supply(Water Quality) Regulations but schedule 1 part 2 national requirementssets a standard for both taste and odour as “acceptable to consumersand no abnormal change”. At least 369 consumers rejected the water fordrinking or washing or cooking (or a combination thereof) based onobjectionable taste and odour. Six of these provided witness statementsto this effect. Accordingly Thames Water Utilities Ltd breached the tasteand odour standard and supplied water that was not wholesome asdefined in Regulation 4(1) and these contraventions were not trivial14.3. The water supplied from Chigwell Works by Essex and Suffolk Water(Northumbrian Water Limited) contravened the Water Supply (WaterQuality) Regulations 2000 (as amended) by virtue of the presence of 2-EDD and 2-EMD, which led to the rejection of the water for drinking,cooking or washing by 43 consumers responding to a questionnaireissued by the Inspectorate and six provided witness statements to thiseffect.14.4. There is no specific standard for these chemicals in the Water Supply(Water Quality) Regulations but schedule 1 part 2 national requirementssets a standard for both taste and odour as “acceptable to consumersand no abnormal change”. . A total of 84 samples were taken fromconsumer taps between 3 February and 19 March 2010. and 15 wereassociated with a positive odour (maximum 19 DN). Accordingly Essexand Suffolk Water (Northumbrian Water Limited) breached the taste andodour standard and supplied water that was not wholesome as definedin Regulation 4(1) and these contraventions were not trivial.36

15. Notification15.1. Thames Water notified the North East and North Central London HealthProtection Unit on 9 February 2010. Essex Health Protection Unit wasnotified on 17 February 2010. Thames Water notified three LocalAuthorities (Tower Hamlets Council, Newham Council and HackneyCouncil) on 10 February 2010. Two Local Authorities were notified on 16February 2010 (Waltham Forest Council and Islington Council) and afurther two had messages left for them on 16 February with a follow upcontact on 17 February 2010 (Epping Forest Council and RedbridgeCouncil). The Consumer Council for Water were notified on 10 February2010.15.2. Some of the notifications were significantly later than others thereforeThames Water did not meet fully the requirements of Section 35(8) ofthe Water Supply (Water Quality) Regulations 2000 (as amended).15.3. Thames Water notified the Inspectorate on 9 February 2010 andprovided relevant reports by the agreed dates however the Company didnot meet fully the notification and reporting requirements of Section 7 ofthe Water Undertakers (Information) Direction 2004. A water suppliermust notify the drinking water quality regulator of the occurrence of anyevent which has caused or, in the opinion of the supplier, is likely tocause, significant concern to persons to whom water is supplied and thismust be given as soon as possible after the event or matter has come tothe supplier‟s attention. The company was aware of a deterioration inquality of the water and took responsive actions at a point in time wellbefore it made the notification.15.4. Essex and Suffolk Water notified the North East and North CentralLondon Health Protection Unit on 11 February 2010. The company alsonotified the London Borough of Barking and Dagenham, LondonBorough of Redbridge and the London Borough of Havering on the sameday. Therefore the company met the requirements of Section 35(6) ofthe Water Supply (Water Quality) Regulations 2000 (as amended).Despite the notification being two days after that of Thames Water,Essex and Suffolk Water notified the authorities as soon as they becameaware of unusual taste and odour reports from consumers.15.5. Essex and Suffolk Water contacted the Inspectorate on 11 February2010 and provided associated reports by the agreed dates and thereforemet the notification and reporting requirements of Section 7 of the WaterUndertakers (Information) Direction 2004 for the same reason asspecified in the preceding paragraph.37

16. Water unfit for human consumption16.1. Water may be regarded as being unfit for human consumption if either,when drunk it would be likely to, or did in fact, cause injury to theconsumer or, where by reason of its appearance or smell, it was of suchquality that it would cause a reasonable consumer of firm character torefuse to drink it or use it in the preparation of food. The supply of waterunfit for human consumption is an offence under Section 70 of the WaterIndustry Act 1991.16.2. Evidence exists that water was rejected by 369 consumers of ThamesWater Utilities Ltd and by 43 consumers of Essex and Suffolk Water(Northumbrian Water Limited) for drinking or washing or cooking, or acombination thereof, based on objectionable taste and odour. Twelveconsumers gave statements to this effect, (six from each company).16.3. Based on the aforementioned evidence that water unfit for humanconsumption may have been supplied during this event the Inspectorateconsidered proceeding with a prosecution under Section 70 of the WaterIndustry Act 1991. Statements were taken from relevant staff of bothcompanies. An interview under caution was carried out with a directingmind of each company. As outlined in this report both companies havetaken number of actions to prevent a recurrence and will be takingfurther actions, including sharing the learning with the industry as awhole and the Consumer Council for Water and other stakeholders suchas local authorities. After careful examination of all available information,including similar historical events worldwide and advice received fromthe Health Protection Agency that neither chemical at the levels foundposed a potential danger to human health, the Chief Inspector ofDrinking Water concluded that proceeding with a prosecution would notbe in the public interest. Each company has admitted the offence ofsupplying water unfit for human consumption during this event and eachhas been formally cautioned. These cautions have been lodged as arecord of the actions that each company has committed to undertake toprevent a recurrence of this incident. The caution record will be takeninto account should at any time in the future a similar event occur. Acopy of the Inspectorate‟s Enforcement Policy is available on its website.16.4. Correspondence relating to this report should be addressed in the firstinstance to Marcus.rink@defra.gsi.gov.uk38

17. References1 The Water Supply (Water Quality) Regulations 20002 The Water Industry Act 19913 The River Severn Pollution Incident of April 1994 and its impact onpublic water supplies. K. J. Ives, D. Hammerton & R. F. PackhamReport for Severn Trent Water Ltd. 11 August 19944 The determination of taste and odour in drinking waters (2010)Methods for the Examination of Waters and Associated MaterialsStanding Committee of AnalystsPublished by the Environment Agency5 Factors causing off-taste in waters, and methods and practices for theremoval of off-taste and its causesFinal Report to the Department of the Environment, Transport andthe RegionsDETR/DWI 5008/1NOVEMBER 20016 Letting the nose lead the way. Malodorous components in drinkingwaterG. Preti, T. S. Gittelman, P. B. Staudite & P. LuitwilerAnalytical Chemistry Vol 65, No. 15, Aug 1, 1993 PP 699 – 7027 Identification of 1,3-Dioxanes and 1,3-Dioxolanes as MalodorousCompounds at Trace Levels in River Water, Groundwater, and TapWaterJ. Romero et alEnvironmental Science and Technology. Vol 32, No 2. 1998 PP 206– 2168 An acute taste and odour episode solved by olfactory GC-MSA. Bruchet, C. Hochereau and C. CamposWater Science & Technology. Vol 55 No 5 PP 223-230 20079 The Worcester water incident, UK: bias in self reported symptoms tothe emergency helplineD. L. Fone. C. E. Constantine, B McCloskeyJ Epidemiol & Community Health 1998; 52: 526 – 52710 An epidemiological study after a water contamination incident nearWorcester, England in April 1994S. E. Fowle, C. E. Constantine, D. Fone, B McCloskeyJ Epidemiol & Community Health 1996; 50: 18-2311 The Upper Lee Catchment Abstraction Management StrategyFinal Strategy Document June 2006Published by the Environment Agency12 The London Catchment Abstraction Management StrategyFinal Strategy Document June 2006Published by the Environment Agency13 UKWIR's Database on Chemical Toxicity, Environmental Fate andWater TreatmentUK Water Industry Research Limited, 1 Queen Anne's Gate,London, SW1H 9BT39

Appendix I: Simplified Schematic of the Raw Water SupplyArrangements40

Appendix II:A: Water Treatment and Supply (Thames Water)RawWaterCoppermills WTWChlorineSulphur DioxideRapid GravityFilersOzoneSlow Sand Filterswith ActivatedCarbonPre-ContactScreensContact TankSupplyOrthophosphoricAcidAmmoniaEpping Forest, Roding Valley, Edmonton, North Tottenham, South Tottenham, Bowes Park, South Enfield,Haringey West, Southgate, Hainault, Highams Park, Woodford, Clapton, Bethnal Green, Mile End,Whitechapel, Isle of Dogs North, Isle of Dogs South, Finsbury Park, Highbury, Pentonville, Bloomsbury,Walthamstow, Leyton, Leytonstone, Wanstead Park, West Ham, East Ham, Plashet grove, Stratford,Hackney West, Hackney East, Hackney South, Stepney, Royal DocksHoddesdon, Goffs Oak, Picketts Lock, Bush Hill, Lee Valley, Enfield Town, UpshireOrthophosphoricAcidSodiumBisulphiteSupplyOzoneContactorsDissolved AirFlotationRapid GravityFilersGranular ActivatedCarbonContactersContact TankSulphuricAcidPoly AluminiumChlorideSodiumHydroxideSodiumHypochloriteAmmoniumSulphateRawWaterChingford South WTW41

B: Water Treatment and Supply (Essex and Suffolk Water)RawWater ChlorineChigwell WTWOzoneSedimentationTanksPrimary FiltersSlow Sand Filterswith ActivatedCarbonContact TankSupplyHarold Hill, Hornchurch, Ilford, Romford East, Romford West, Seven Kings, Thames,Barkingside42

Appendix IIIA: Abstraction from the River Lee to King George V, William Girling andHigh Maynard Reservoirs, (ML/day)43

B: Percentage contribution of the raw water supply from King GeorgeV, William Girling and the Thames Lee Tunnel to Coppermills WTW44

Appendix IVA: Comparison of the total number of contacts received withconcentrations of 2-EMD and 2-EDD encountered at a zonal samplingpoint (Thames Water Woodford Depot).Graph provided by Thames Water.45

B: Contacts received by Essex and Suffolk water during the eventGraph derived from Essex and Suffolk Water data46

Appendix V:A: Consumer contact distribution map for ThamesWater47

B: Consumer contact distribution map for Essex and Suffolk Water48

Appendix VI:A: Method synopsis of 2-EMD/2-EDDSupplied by Thames Water LaboratoryEach 500ml sample is acidified with sulphuric acid to pH2 and ascorbic acidadded to remove any residual chlorine. The sample is then spiked with D5Chlorobenzene (at 200 ng/l), which is used as a surrogate standard.A sample is extracted with 20 ml of dichloromethane, by shaking for 4minutes at 190 rpm on a linear flat bed shaker, the dichloromethane layer isremoved and the extraction repeated with a 10ml aliquot of dichloromethane.The extracts are dried with anhydrous sodium sulphate, and then combined.The combined extract is then concentrated using a combination of a waterbath and nitrogen gas, until this extract volume is reduced to 200 µl.With each batch of samples a procedural blank, a set of calibration standards(10, 50, 100 & 200 ng/l), and an AQC are also analysed.Samples are run on an Agilent 7890A GC coupled to an Agilent 5975C MS.Samples introduction was, via an automated sampler, into a cool on-columninjector. A 1m length of 0.53mm id deactivated fused silica pre-column wasconnected to a 30m x 0.25mm id x 0.25 µm film thickness DB-5ms capillarycolumn with a constant flow rate of 1ml/min of Helium.GC Conditions:Initial temperature: 35 °C; hold for 4.0 minutes,Temperature Ramp 1: 10 °C/min to 80 °C,Temperature Ramp 2: 50 °C/min to 300 °C; hold for 10.0 minutes.Injection volume: 2 µlMass spectral detection is carried out in selected ion monitoring (SIM) modewith the following ion used for each compound.CompoundQuantitative ion(m/z)Qual Ion1 (m/z)Qual Ion2 (m/z)QualIon 3(m/z)D5 Chlorobenzene 117 1192-EMD 87 59 72 1152-EDD 115 56 69 143MS Interface temperature: 310 °C.49

B: Mass Spectra for 2-EDD(Actual 18 Feb 2010 produced by Thames Water Laboratory)50

C: Mass Spectra for 2-EMD(Actual 18 Feb 2010 produced by Thames Water Laboratory)51

Appendix VII: Technical details of 2-EDD & 2-EMDInformation reproduced with the kind permission of UKWIR 12A: 2-Ethyl-5,5-dimethyl-1,3-dioxane (2-EDD)Synopsis2-Ethyl-5,5-dimethyl-1,3-dioxane (2-EDD) has been identified in waste products from a resinfactory. It has been involved in several incidents involving contamination of drinking water andcaused major taste and odour problems. However, an epidemiological study conducted afterone of these incidents failed to demonstrate any adverse health effects as a result of thechemical contamination, although adverse symptoms were associated with the unpleasanttaste or odour of the tap water. 2-EDD has the ability to pass through sewage treatment andwas not removed by coagulation, flocculation or clarification during drinking water treatment.The use of granular activated carbon (GAC) reduced the concentration of the contaminantand powdered activated carbon (PAC) was rapid and effective in removing 2-EDD. Therewere no data on the toxicity of 2-EDD, although it is likely to cause odour problems atsignificantly lower levels than those considered to be of risk to human health. It is extremelyodorous and an operational SNARL of 0.01 µg/l (10 ng/l) is suggested based on odour,although some sensitive individuals may be able to detect odour as low as 0.005 µg/l (5 ng/l).OccurrencesIn January 1992, a commercial hazardous waste management facility, located 28 milesupstream of a water treatment plant accepted six tanker loads (about 30 000 gallons; 113 562litres) of wastewater from a resin coatings manufacturer in New Jersey, USA. The watercontained substantial concentrations of various by-products of the resin-manufacturingprocess. One of these by-products proved to be a very potent odour-causing agent that had abitter taste. Because this material was not removed by treatment at the waste managementfacility or at the sewer plant to which the treated wastewater was discharged, the sewer plantand the creek were subsequently contaminated with the odour-causing agent (1). Odourcomparisons, smell chromatography and GC/MS analyses showed that 2-EDD was thepredominant odour-causing agent in the waste associated with the incident (1). Estimates of2-EDD concentrations during the height of the incident, based on comparisons of watersamples with measured concentrations of the synthetic 2-EDD are provided in the tablebelow:Location52Concentration(µg/l)Wastewater from the resin coating manufacturer 950Effluent from a sewer plant 10Water from the west branch of the creek(approximately 300 feet downstream of thesewerage plant)Water from the abstraction point 0.04Finished drinking water 0.02A similar incident occurred in the UK in April 1994. On 11th April 1994, a mixture of organicchemicals was discharged into a sewer, passed through Wem sewage treatment works andentered the River Severn via two tributaries. Four days later in Worcester (about 120 kmdown river from Wem), water consumers complained of an unpleasant taste and odour.Analysis of a sample from Wem (taken on 13 April) showed a mixture of 7 organics and 2-EDD was detected at a level of 1300 mg/l. By 17 April, levels of 2-EDD had fallen toanalytically undetectable concentrations. The highest concentration measured in treatedwater was 0.13 µg/l (2).5

In an investigation of taste and odour episodes involving Barcelona's drinking water supplybetween 1993 and 1995, 2-EDD was detected in wastewater from a resin plant and ingroundwater samples (distance 0-40 km). The concentration of 2-EDD found in wastewaterwas 3673 µg/l and, in groundwater samples, ranged between below the level of detection to7.07 µg/l (mean concentration: 0.95 µg/l) (7).In 2010 in the UK, following a series of consumer taste and odour complaints, 2-EDD wasdetected in a drinking water supply at a concentration range of 0.03-0.2 µg/l (WRc data).UsesNo data were located at time of update. However, 2-EDD has been identified as a by-productof the resin manufacturing process (type of resin not reported).Human toxicityNo specific health effects are known for this chemical. However, an epidemiology study wasperformed to assess the health effects of the Worcester incident that occurred in 1994 (10).This study found that 62% of people questioned noted an unusual taste or odour in the waterand that a dose-response relationship existed between the amount of water consumed andthe presence of a variety of symptoms (diarrhoea, nausea, headache, stomach pains, skinirritation and itchy eyes) (10). However, the study concluded that the symptoms presentedwere associated with the unpleasant taste or odour of the tap water rather than the chemicalcontamination (10).Emergencies and SNARLs2-EDD is extremely odorous with a variety of odour thresholds being reported between 0.01-0.02 µg/l, although sensitive individuals may be able to detect odour as low as 0.005 µg/l. AWRc taste threshold of 0.02 µg/l has also been reported. An operational SNARL of 0.01 µg/l(10 ng/l) is recommended in order to avoid odour effects.Taste and odour propertiesBased on the literature, the odour threshold for 2-EDD in water appears to be 0.01 µg/l (1).The odour has been described as walnutty, latex paint, varnish, chlorinous, earthy, musty,creeky/decaying vegetation, nutty, sewage, fishy/algae, marshy/sulphurous, rotten,methanol/piney, sweet, chemical (4).In a study conducted in 1997 (5) using a sniff port on a gas chromatograph the EDD peakwas described as tutti-frutti (Italian for 'all fruits', but implying sweet, fruity confectionery). Thestudy included Flavour Profile Analyses which produced descriptors such as fruity, apple,menthol, sweet, burnt-sweet, solvent-sweet and sickening-sweet and suggested an odourthreshold concentration of 0.005 to 0.01 µg/l. At levels close to the threshold concentrationthe odour was described using the more pleasant sweet terms, but less pleasant descriptorswere applied to slightly more concentrated solutions.WRc Data:Tests were carried out on solutions of 2-EDD (distilled and pure by GC analysis) dissolved instill mineral water. Odour tests were done at 40°C and taste tests at 25°C. The test panelsconsisted of ten assessors but in each case one member failed to produce consistent resultsand therefore the following information summary is based on the results of the other nineassessors.The minimum concentrations detected were odour at 6 ng/l and taste at 16 ng/l and allassessors detected both an odour and taste in the 100 ng/l samples. The median thresholdconcentrations were calculated as 17 ng/l for odour and 22 ng/l for taste.The descriptors were very varied, but is was noted that the odour was more objectionable atthe lower concentrations. During the preparation of the test solutions the odour of significantlymore concentrated solutions was perceived as minty and menthol. These observations are in53

agreement with those made when testing the odour of atmospheric samples (6). The odourwas described as chemical/paint, chemical, sweet, stale, acetate, musty, effluent, plastic,yeast, estery, car-polish, perfumed, cowsheds, paint, rotting, fruit, plasticine and dishwater.The taste was described as metallic, musty, chemical, plastic, stale, acetate, fruity, sweet,rotting vegetables, cowsheds, drains, salty, burning, plasticine, vinegar, estery, car-polish,bitter, rotting apples and mouldy.Removal during (drinking water) treatmentCoagulation, flocculation and clarification do not appear to remove 2-EDD. However, the useof Granular Activated Carbon (GAC) and Powdered Activated Carbon (PAC) are effective inremoving 2-EDD (2).54

B: 2-Ethyl-4-methyl-1,3-dioxolane (2-EMD)Synopsis2-Ethyl-4-methyl-1,3-dioxolane (2-EMD) has been identified in waste products from a resinfactory. It has been involved in several incidents involving contamination of drinking water andmajor taste and odour problems. However, 2-EMD was not considered to be the odouroussource, and an epidemiological study conducted after one of these incidents failed todemonstrate any adverse health effects as a result of the chemical contamination.Coagulation, flocculation and clarification do not appear to remove 2-EMD. However, the useof Granular Activated Carbon (GAC) and Powdered Activated Carbon (PAC) are effective inremoving 2-EMD. No data were located on the toxicity of 2-EMD, but based on a structurallysimilar compound, it is likely to be of low acute oral toxicity to experimental animals. Instead, itis likely to cause odour problems at much lower levels than those considered to be of risk tohealth in the short-term. Odour threshold values reported in the literature vary considerablyfrom as low as 0.005 to 380 µg/l, the odour most often being described as 'sweet'. A tentative24-hour health-based SNARL of 1400 µg/l is proposed. Given the variation in thresholdvalues it is difficult to propose an operational SNARL based on odour. It will be important tomonitor taste and odour should 2-EMD contaminate water sources or supply.OccurrencesCocheo et al. (1) reported that 2-EMD was found in the effluent from the sewage treatmentplant of a fibreglass and resin manufacturing company (levels not reported) as a by-product ofthe formation of propionaldehyde. Since that time, 2-EMD has been detected as a by-productpresent in waste from resin manufacturing plants and has been involved in a number ofpollution incidents (2, 6, 7).2-EMD was detected in a pollution incident on the River Severn in April 1994. On 11th April1994, a mixture of organic chemicals was discharged into a sewer, passed through Wemsewage treatment works and entered the River Severn via two tributaries. Four days later inWorcester (about 120 km down river from Wem) water consumers complained of anunpleasant taste and odour. Analysis of a sample from Wem (taken on 13 April) showed amixture of 7 organics and 2-EMD was detected at 350 mg/l. By 17 April levels of 2-EMD hadfallen to analytically undetectable concentrations. The highest concentration measured intreated water was 0.02 µg/l (2). However, 2-EMD was not thought to be the source of odour.Instead, the compound causing the taste and odour problems was identified as 2-ethyl-5,5-dimethyl-1,3-dioxane (2-EDD) (2). A Toxicity Datasheet is available for 2-Ethyl-5,5-dimethyl-1,3-dioxane.In a taste and odour incident along the Ohio river in the USA in 1989, 2-EMD was detected ina discharge from a sewage treatment plant (levels not reported). Like the Worcester incident,taste and odour effects were detected downstream (as far as 137 miles). Due to itshydrophilic nature, 2-EMD was not sufficiently removed by the wastewater treatment plantusing activated sludge treatment nor by drinking water treatment facilities downstream (6).In an investigation of groundwater contamination involving Barcelona's water supply, 2-EMDwas detected in wastewater from a resin plant and in groundwater samples (distance 0-40km). The concentration of 2-EMD found in waste water was 585 µg/l and in groundwatersamples, ranged between below the level of detection to 30.6 µg/l (mean concentration: 4.02µg/l) (7).UsesNo data on the use of 2-EMD were located at time of update. However, it has been identifiedas a by-product of the resin manufacturing process (type of resin not stated).Human toxicityNo specific health effects are available for this chemical. However, an epidemiology studywas performed to assess the health effects of the Worcester incident that occurred in 1994(11). This study found that 62% of people questioned noted an unusual taste or odour in the55

water and that a dose-response relationship existed between the amount of water consumedand the presence of symptoms (diarrhoea, nausea, headache, stomach pains, skin irritationand itchy eyes) (11). However, the study concluded that the symptoms presented wereassociated with the unpleasant taste or odour of the tap water rather than the chemicalcontamination (11).Emergencies and SNARLs2-EDD is extremely odorous with a variety of odour thresholds being reported between 0.01-0.02 µg/l, although sensitive individuals may be able to detect odour as low as 0.005 µg/l. AWRc taste threshold of 0.02 µg/l has also been reported. An operational SNARL of 0.01 µg/l(10 ng/l) is recommended in order to avoid odour effectsTaste and odour propertiesThe odour threshold of 2-EMD was found to be 380 µg/l at room temperature and 360 µg/l at60°C based on a panel size of 8 (5). The range of odour thresholds detected were 140-1390µg/l and 70-810 µg/l at room temperature and 60°C, respectively. The odour was describedas 'terpene', 'sweet' or 'spicy'.A recent abstract reported in the proceedings of a conference (6) has reached differentconclusions and reported a much lower odour threshold. From Flavour Profile Analysesusing a trained panel the odour threshold concentration was estimated to be 0.005 to 0.01µg/l, with descriptors being 'sweet' (at threshold concentration) and, at higher concentrations,'solvent sweet', 'sickly sweet', 'toluene' and 'medicinal'. Less experienced panellists used thedescriptors 'solvent', 'toluene', 'phenolic', 'sweet', 'solvent sweet' and 'sickening sweet'. Fromthe method of preparation it must be assumed that the material tested was a mixture of thefour stereoisomers.56

Appendix VIII: Customer Questionnaire Responses to DWI1. Did you notice anything different about yourwater supply around February 2010?Essex & SuffolkThames WaterWaterNo‟ % No‟ %Yes 46 96 394 98No 2 47 2Total 48 4012. When did you first notice that there wassomething different about your drinking water?% %Dec 2010 2 7 15 6Jan-10 7 24 87 32Feb-10 20 69 163 61Mar-10 0 0 3 1Total 29 2683. Did you notice the difference by:% %Smell 1 2 15 4Taste 8 17 59 15Smell & Taste 32 68 263 67Appearance 1 2 0 0All three 5 11 54 14Other 0 0 4 1Total 15 3915. Did the change in the water supply causeyou to stop using it for any of the following:% %Drinking 26 60 213 58Cooking 0 0 1 0.3Drinking & Cooking 13 30 114 31Drinking & Washing 0 0 8 2Cooking & Washing 0 0 1 0.3Drinking, Cooking & Washing 2 5 15 4Cooking, Washing, Laundry 1 2.5 0 0Drinking, Cooking, Washing, Laundry 1 2.5 12 3Other combination 0 0 5 1.3Total 43 3696. Approximately how long did the problemcontinue?% %1 day 0 0 2 12-7 days 7 5 19 58-30 days 30 70 199 55>30 days 11 26 145 40Total 43 36557

7. Did you receive any advice from your watercompany?% %Contact in advance 0 0 0 0Contacted after problem 0 0 7 2Only when I contacted Company 32 73 303 78No advice 12 27 78 20Total 44 3889. Did the Water Company offer you analternative supply of water?Bottled 0 0 5 1Tanker or Bowser 0 0 0 0Other 0 0 8 2No 45 100 379 97Total 45 39210. Did you take a sample of the water, if so,what happened to it?% %Still have it 0 0 6 2Gave to Water Company 0 0 6 2Gave to Env Health 1 2 1 0Threw Away 2 4 27 7Other 1 2 9 2Didn't take Samples 41 91 341 87Total 45 39011. Did the Water Company take a samplefrom your property?% %Yes Got results 3 7 24 6Yes didn't get results 1 2 4 1No 41 91 358 93Total 45 38612. Did you hear about the problem in the localor national media?Radio Only 0 0 5 2TV Only 11 46 118 44Newspaper Only 3 13 59 22Radio & TV 2 8 14 5Radio &Newspaper 0 0 2 1TV & Newspaper 6 25 49 18Radio, TV, & Newspaper 1 4 13 5Other 1 4 6 2None 0 0 0 0Total 24 26658

Appendix IX: BBC Website reports of eventReproduced from the BBC Websitehttp://news.bbc.co.uk/1/hi/england/london/8517463.stm andhttp://news.bbc.co.uk/1/hi/england/london/8539286.stmTuesday, 16 February 2010Thames investigate 'smelly water' inWalthamstowReports of foul smelling drinking water are beinginvestigated after hundreds of complaints were madeby residents in Walthamstow, north-east London .Thames Water has stopped taking supplies from the RiverLea and is now extracting water from the River Thames whilethey consider the issues.Water tests have been carried out but so far nothingabnormal has been found.Since the start of February, 300 of the 1.8m customers served by the supply inWalthamstow have reported the problem.Bob Collington, Director of Operational Management for Thames Water, said: "Earlyindications suggest the problem could be due to a change in environmental conditions in theRiver Lee, which is the normal source of water for our Walthamstow works."We've spoken to the Health Protection Agency for the area and our drinking waterregulator, the Drinking Water Inspectorate, and we can confirm that they are supportive ofthe steps we're taking."As well as carrying out extra testing at our own laboratories, we've also sent samples tolabs at other water companies and to independent labs, but so far nothing abnormal hasbeen identified."Thames Water supplies water to over eight million customers across London and theThames Valley.Since the start of February 300people have complained59

Friday, 26 February 2010Chemicals caused smelly tap water in northeastLondonFoul smelling drinking water reported by residentsof east and north London was caused by a chemicalused in glues and manufacturing, it has emerged.Since the start of February 800 people have complainedabout the smell, caused by a contamination of the RiverLea.Now Thames Water has said the smell was caused by thechemicals 2-EDD and 2-EMD, present at harmless levels.At its height there were 30 billionths of a gram of the chemicals in each litre of water.A Thames Water spokesman said the chemical had a "low smell threshold", meaningsome people could detect it even at very low levels.He said: "Our most recent tests indicate that minute traces of these substances in watergoing to customers in north-east London have significantly reduced and are now almostundetectable.'We're really sorry'"Although 800 of the two-million customers served by our Walthamstow works havereported an unusual smell in their water since the start of February, this has never beena health concern."We're really sorry to all those who've been affected."The Drinking Water Inspectorate has begun an investigation.A spokeswoman confirmed that if the water was found to be unfit for consumption orinadequately treated Thames Water could face fines of thousands of pounds.She said: "When ongoing investigations are complete the inspectorate will independentlyassess all the findings and then report on the cause and the lessons learnt."Consumers experiencing ongoing taste and odour problems should contact their watercompany in the first instance."Since the start of February 800people have complained60